Practical Perl Tools Oh Say Can You CPAN?
Total Page：16
File Type：pdf, Size：1020Kb
Practical Perl Tools Oh Say Can You CPAN? COLUMNSDAVID N. BLANK-EDELMAN David N. Blank-Edelman is very once in a while I get asked to join a conference panel about the Director of Technology at scripting languages. It will be me, a Pythonista, a Rubyist, and a the Northeastern University PHP developer (do they have a cute name?) all onstage together. In College of Computer and E Information Science and the most cases, I think the organizers are hoping for the equivalent of a steel author of the O’Reilly book Automating System cage match in professional wrestling—the fewer participants standing at Administration with Perl (the second edition of the end, the better. In these scenarios, I’m almost always a disappointment the Otter book), available at purveyors of fine because I come to praise the other languages, not to bury them. I have a deep dead trees everywhere. He has spent the past appreciation for the other languages, and I’m not afraid to state it even while 24+ years as a system/network administrator I’m representing Perl. One of the key reasons I can say “I dig all of the other in large multi-platform environments, including Brandeis University, Cambridge Technology languages, but I choose to stick with Perl most of the time” is CPAN. This col- Group, and the MIT Media Laboratory. He was umn will focus on CPAN, how to cope with both its triumphs and shortfalls, the program chair of the LISA ‘05 conference and some of the ways to interact with it that you may not have encountered and one of the LISA ‘06 Invited Talks co-chairs. before. There probably won’t be any code in this issue’s column but that’s David is honored to have been the recipient okay because you’ll be learning ways to have other people write Perl code for of the 2009 SAGE Outstanding Achievement you. We’re going to focus on how to consume content from CPAN; discussion Award and to serve on the USENIX Board of Directors beginning in June of 2010. about how to contribute to it will have to wait for a future column. [email protected] What Is CPAN and How Do I Get Me Some? I would be really surprised if there are Perl programmers who have never heard of CPAN, but I’ve been surprised before so pardon me as I go over the basics. CPAN is short for the Com- prehensive Perl Archive Network. This is a massive repository of Perl code (largely modules meant for use in other people’s code) that has been online since 1995 or so. How massive? As of this writing, cpan.org says: The Comprehensive Perl Archive Network (CPAN) currently has 138,392 Perl modules in 30,406 distributions, written by 11,739 authors, mirrored on 254 servers. All of this code has been uploaded so other people may make use of it, so it can be a tremen- dous resource. Any time you have a problem or a task that sounds like someone else may have solved it, it always behooves you to search CPAN first. We’ll talk about ways to do this in a moment. The plus of having such a massive store of donated code to draw upon is that you often can find someone else has already written (almost) exactly what you need. The minus of having this massive store is some percentage of it is (to be charitable) duplicated effort, and (to be less charitable) some of it is crap. I’ll offer tips about this problem later in this column as well. Deep CPAN Diving So how do you find what is available on CPAN? Many people start with the search.cpan.org engine. This Googley-looking search engine returns a page like the one in Figure 1. An experienced CPAN spelunker will scan the returned list of modules and look not just at the description to determine whether a module is appropriate for the task at hand, but also at the metadata. For example, has the module been updated recently? Does it have any reviews 42 DECEMBER 2014 VOL. 39, NO. 6 www.usenix.org COLUMNS Practical Perl Tools: Oh Say Can You CPAN? Figure 1: A screenshot of search.cpan.org showing some results and are they positive? Is the module part of a distribution I recognize? Is the module author well known in the Perl commu- nity?, etc. When you click on the module name, you’ll be greeted with the documentation for that module. More often than not, I will click on the breadcrumb link that brings me to the page for the whole distribution (Figure 2). I do this for two reasons: First, I often want to poke around Figure 2: Clicking on the breadcrumb link brings up the page for the whole in a module’s code (especially looking at the test code in it for distribution. examples of how to use the module). This can be done from the Browse link. Second, I might be curious about bugs filed against the module (“View/Report Bugs”) or what other modules this module depends on (“Dependencies”—we’ll talk more about that soon). Some of these links can be reached from the search results or the first page linked off the search results, but I’m so used to using the Browse link that going to the distribution page is habitual at this point. Another way you can search for modules on CPAN is to use metacpan.org. MetaCPAN attempts to be an even spiffier search engine. Figure 3 shows the same search from before, this time run at MetaCPAN.org. First, let’s talk about what is spiffier on the service. When I first started typing “Readonly” into the search box, it attempted to auto-complete my query. Next, not only are Readonly and Readonly::XS next to each other, but at the bottom of the results you can see MetaCPAN has bunched together related modules in a distribution. When I click on the first module link on this page, I see the page that begins like Figure 4. Figure 3: Using MetaCPAN as your search engine www.usenix.org DECEMBER 2014 VOL. 39, NO. 6 43 COLUMNS Practical Perl Tools: Oh Say Can You CPAN? (Launchbar) to make looking up module documentation quick and easy: https://metacpan.org/search?q=* Separating the Wheat from the Chaff Now that you know how to find more modules than you can shake a stick at, how do you figure out which are the good ones? I’ve mentioned a couple of ideas already in passing, but let’s take a closer look at this question. First, I think it is worthwhile to favor modules that appear to be actively maintained. This increases the likelihood that there is an author out there improving the module and also available to fix issues should you find any. The last release date is a good hint about this, the activity indictor provided by metacpan.org is an even better indication. Second, closely related to the first idea is the number of bugs opened against the module. I don’t believe zero active bugs is necessarily a good thing. I’d much rather see a few open bugs (shows community involvement) alongside a number of closed bugs (shows author involvement and responsiveness). A queue full of unresolved bugs is also a great red flag that may indicate an orphaned module. Use this as one of your parameters for judg- ment but not the only one. Third, consider whether the module appears to actually work. One way to determine this is to look at the Testers link off of the metacpan module page for a module. To return to the panels I mentioned at the beginning of the column, another thing I believe Perl can be proud about is the strong cultural inclination towards testing in the community. One way this manifests is that every version of every module that gets submitted to CPAN Figure 4: Clicking on the first module link after searching at MetaCPAN gets “smoke tested” on close to a thousand different combina- I’d like to call your attention to a few of the things in the left tions of Perl versions and operating systems. If a module includes and right sidebars. On the left sidebar, I have the “Browse” link tests (and indeed, every module is encouraged to have as com- I crave, an easy way to look at the Changelog, pointers to the Git plete a test suite as possible), these tests are run in each of these repo, the Bug tracker page for the module, reviews, test results environments and the results reported back to the central test (more on this later), a link to an automated system for determin- result repository for you to peruse. This gives you a good indica- ing module quality (“Kwalitee”), an indication how active the tion (again, if the test suite is decent) of how portable and how development is of the module, and even a way to download the fragile the module code is likely to be. I’ve mentioned a couple module doc in several ebook formats plus a bunch more stuff. of times that a good test suite makes this metric useful, and I’d On the right sidebar, we see the dependencies of the module and, recommend using the Browse link to see what sort of tests are perhaps even cooler, a way to see the reverse dependencies (i.e., included with a module.